Prepared for submission to JCAP

Out of the White Hole:

A Holographic Origin for the Big Bang
Razieh Pourhasan,
a,b
Niayesh Afshordi,
a,b
and Robert B. Mann
a,b
a
Department of Physics & Astronomy, University of Waterloo, Waterloo, Ontario N2L 3G1,
Canada
b
Perimeter Institute for Theoretical Physics, 31 Caroline St. N., Waterloo, ON, N2L 2Y5,
Canada
E-mail: r2pourhasan@uwaterloo.ca, nafshordi@pitp.ca, rbmann@uwaterloo.ca
Abstract. While most of the singularities of General Relativity are expected to be safely
hidden behind event horizons by the cosmic censorship conjecture, we happen to live in the
causal future of the classical big bang singularity, whose resolution constitutes the active
eld of early universe cosmology. Could the big bang be also hidden behind a causal horizon,
making us immune to the decadent impacts of a naked singularity? We describe a braneworld
description of cosmology with both 4d induced and 5d bulk gravity (otherwise known as Dvali-
Gabadadze-Porati, or DGP model), which exhibits this feature: The universe emerges as a
spherical 3-brane out of the formation of a 5d Schwarzschild black hole. In particular, we
show that a pressure singularity of the holographic uid, discovered earlier, happens inside
the white hole horizon, and thus need not be real or imply any pathology. Furthermore,
we outline a novel mechanism through which any thermal atmosphere for the brane, with
comoving temperature of 20% of the 5D Planck mass can induce scale-invariant primordial
curvature perturbations on the brane, circumventing the need for a separate process (such
as cosmic ination) to explain current cosmological observations. Finally, we note that 5D
space-time is asymptotically at, and thus potentially allows an S-matrix or (after minor
modications) AdS/CFT description of the cosmological big bang.
Keywords: gravity, cosmology, big bang singularity, braneworld scenarios, early universe
ArXiv ePrint: 1309.xxxx
a
r
X
i
v
:
1
3
0
9
.
1
4
8
7
v
2


[
h
e
p
-
t
h
]


2
2

S
e
p

2
0
1
3
Contents
1 Introduction 1
2 Universe with FRW metric 2
3 Universe as a hologram for a Schwarzschild bulk 4
4 Brane Atmosphere and Cosmological Perturbations 9
5 Summary and Discussions 12
1 Introduction
The scientic discipline of Physical Cosmology started as, and continues to be, an extremely
ambitious attempt to summarize the physics of the entire universe within a handful of cosmo-
logical parameters. However, maybe the most surprising outcome of this enterprise has been
how successful this naive approach has been in describing cosmological observations that are
multiplying at an accelerating rate. This is exemplied by the spectacular data recently re-
leased by the Planck collaboration [1], and its remarkable agreement with the six-parameter
CDM paradigm. However, the experimental success of standard cosmology is overshadowed
by fundamental existential questions: What is Dark Matter? Why Dark Energy? What is
the nature of the Big Bang?
The starting point for this paper was to ask whether a more satisfactory (or natural)
understanding of these mysteries can come from an alternative description of the geometry. In
particular, could these (seemingly unrelated) phenomena be manifestations of hidden spatial
dimensions, that show up as holographic uid(s) in our 4D description?
Motivated by D-branes in 10D string theory, pure phenomenology, or a combination of
the two, one way to describe our four-dimensional universe is through embedding it in a higher
dimensional spacetime with at least one more dimension and investigate its gravitational
and/or cosmological properties. This is known as the brane world scenario, where the
brane refers to our 4D universe embedded in a bulk space-time with 5 or more dimensions,
where only gravitational forces dare to venture. Well-known (and well-studied) examples
of such scenarios are the Randall-Sundrum (RS) [2] model, where 4D gravity is recovered
through a compact volume bulk, or the Dvali-Gabadadze-Porrati (DGP) construction [3, 4],
where our 3-brane is equipped with its own induced gravity, competing with the bulk gravity
via the so-called Vainshtein mechanism [5].
Radiation dominated cosmology has been studied in the context of RS model where
FRW metric describing 4D universe emerges as induced gravity on the brane in 5D AdS/Shchwarzschild
background, e.g. [6, 7]. However in this paper, we focus on the DGP model, which is dened
by the following action:
S
DGP

1
16G
b
_
bulk
d
5
x

gR
5
+
1
8G
b
_
brane
d
4
x

K+
_
brane
d
4
x

_
R
4
16G
N
+/
matter
_
,
(1.1)
where g and are the bulk and brane metrics respectively, while K and R
4
are the mean
extrinsic and Ricci intrinsic curvatures of the brane. G
b
and G
N
are then respectively the bulk
1
and brane (i.e. Newtons) gravitational constants. One may also express the gravitational
constants in terms of the bulk and brane Planck masses:
M
4
= (16G
N
)
1/2
M
5
= (32G
b
)
1/3
, (1.2)
which respectively describe the approximate energies at which the brane and bulk gravitons
become strongly coupled. Moreover, the ratio r
c
G
b
/G
N
characterizes the length scale
above where 5D gravity becomes important.
Along with a great deal of attention, these models have received some criticism. The
DGP model includes a de Sitter solution automatically, which is usually called a self acceler-
ating (SA) branch. When rst proposed, this gave rise to the hope of a consistent description
of our accelerating universe without a cosmological constant. However, it turned out that
SA solutions suer from ghosts and tachyons [811] as well as some pathological singularities
[12]. Furthermore, the detailed predictions of the SA branch were inconsistent with cosmo-
logical observations [13]. Nevertheless, the normal (non-self-accelerating) branch of the DGP
cosmology does not suer from the same pathologies, and can be consistent with data, if one
includes brane tension (which is the same as a 4D cosmological constant) [14].
While most studies in the context of DGP have been made from the viewpoint of a 4D
observer living on the brane, the DGP model was reexamined [15] as a theory of 5D Einstein
gravity coupled to 4D DGP branes, using a Hamiltonian analysis. New pathologies were
encountered in the model by generalizing the 5D geometry from Minkowski space-time as
originally considered in the DGP model to Schwarzschild. If the black hole mass in the
bulk exceeds a critical value, a so called pressure singularity will arise at nite radius [15].
Furthermore, on the SA branch the ve-dimensional energy is unbounded from below.
Here we study the DGP model around a 5D black hole in greater detail to better
understand its phenomenological viability. We relate bulk, brane, and black hole parameters
and investigate constraints on them that allow one to avoid the pressure singularity. We nd
that viable solutions are indeed possible, leading us to propose a holographic description for
the big bang, that avoids the big bang singularity. We further outline a novel mechanism
through which the branes atmosphere induces (near) scale-variant curvature perturbations
on the brane, without any strong ne tuning (or need for additional processes, such as cosmic
ination), consistent with cosmic microwave background observations.
In Section 2, we introduce the induced gravity on the brane by solving the vacuum Ein-
stein equations while we demand a Freedman-Robertson-Walker (FRW) metric on the brane.
In Section 3, we describe the geometry in the bulk in more detail and clarify the holographic
picture of the brane from the point of view of the 5D observer. We then give our proposal
for a holographic big bang as emergence from a collapsing 5D black hole. Section 4 outlines
a mechanism to generate cosmological curvature perturbations from thermal uctuations in
the brane atmosphere. Finally, Section 5 wraps up the paper with a summary of our results
and related discussions.
2 Universe with FRW metric
We start by introducing the standard form of the FRW line element:
ds
2
= d
2
+
a
2
()
/
_
d
2
+ sin
2
_
d
2
+ sin
2
d
2
_
, (2.1)
where / > 0 is the curvature parameter whose dimensions are (length)
2
and the scale factor
a is dimensionless and normalized to unity at the present time, i.e. a(
0
) a
0
= 1.
2
Using the metric (2.1) for the brane, we next turn to solving the Einstein equations on
the brane
G

= 8G
N
(T

+

T

) , (2.2)
where G
N
is the gravitational constant on the brane. We here include two types of energy-
momentum tensor T

and

T

. The former describes normal matter living on the brane in

a form of a perfect uid
T

= (P + )u

+ Pg

,
satisfying the continuity equation

= 0, (2.3)
where g

is the metric on the brane given by (2.1) and u

is the 4-velocity of the uid

normalized such that u

= 1. The latter stress-energy

T

is the Brown-York stress

tensor [16] induced on the brane, dened from the extrinsic curvature K

as

1
8G
b
(Kg

) , (2.4)
where G
b
is the gravitational constant in the bulk, and we have assumed Z
2
bulk boundary
conditions on the brane. The vacuum Einstein equations in the bulk impose the following
constraints on the brane

(Kg

) = 0 , (2.5)
R + K

K
2
= 0 , (2.6)
where R = 8G
N
(T +

T) is the Ricci scalar on the brane. The rst constraint is just the
continuity equation for

T

while the second one is the so called Hamiltonian constraint.

Without loss of generality, as a result of the symmetry of FRW space-time, we can write

in a perfect uid form i.e.

= (

P + )u

+

Pg

, (2.7)
which we shall refer to as the induced (or holographic) uid. Combining Eqs. (2.4) and (2.7),
we get:
K

= 8G
b
_
(

P + )u

+
1
3
g

_
. (2.8)
From Eqs. (2.2), (2.3), (2.5) and (2.6) we respectively obtain
H
2
+
/
a
2
=
8G
N
3
( + ) , (2.9)
+ 3H( + P) = 0 , (2.10)

+ 3H( +

P) = 0 , (2.11)
+ 3(P +

P) +
8G
2
b
G
N
_
2
3

2
+ 2

P
_
= 0, (2.12)
where the last equation follows from solving for K

in terms of ( ,

P) using Eq. (2.8).
3
Combining (2.9-2.12) we get for and

P:

=
s
_
1

1

2
12G
N

s
1
a
4
+
2

s
_
, (2.13)

P =

2
+
s
( T)
3(
s
)
, (2.14)
where
T = 3P (2.15)
and we choose the constant of integration
2
, of dimension [length]
2
, to be negative (see
e.g. [17], for a similar derivation of DGP cosmology). The choice of minus sign will be
justied in the next section, where we introduce the holographic picture. Finally, we have
also dened the characteristic density scale for the holographic uid:

s

3G
N
16G
2
b
. (2.16)
Equation (2.14) immediately implies that the pressure becomes singular at =
s
. It
is then of interest to investigate in whether this pressure singularity can happen at early or
late times (if at all), in our cosmic history. We address this question in the next section.
Furthermore, we note that
s
sets the characteristic density scale, below which the bulk
gravity becomes important. Specically, it is easy to see that both terms in the induced uid
density, (Eq. 2.13), become much smaller than the matter density, , if
s
. Therefore,
given the current lack of observational evidence for 5D gravity (e.g. [14]), it is safe to assume
that (z) >
now

s
, i.e. the induced uid has always had a negligible contribution to
cosmic expansion, with the notable (possible) exception of the above-mentioned singularity.
3 Universe as a hologram for a Schwarzschild bulk
Consider our universe to be a (3+1)-dimensional holographic image [18] call it a brane
of a (4+1)-dimensional background Schwarzchild geometry
ds
2
bulk
= f(r)dt
2
+
dr
2
f(r)
+ r
2
d
2
3
, (3.1)
with
f(r) = 1
r
2
h
r
2
, (3.2)
and where d
3
is the metric of unit 3-sphere. We now assume a dynamical brane, i.e our
universe, to be located at r = a()/

/ described by the FRW metric (2.1). Its unit normal

vector is
n

=
_
a

/f(a)
,
_
f(a) +
a
2
/
, 0, 0, 0
_
, (3.3)
with n

= 1 and = 1 or +1, and we take

u

=
_
1
f(a)
_
f(a) +
a
2
/
,
a

/
, 0, 0, 0
_
(3.4)
4
to be the unit timelike tangent vector on the brane, i.e u

= 1.
Recall that besides normal matter on the brane we also introduced an induced uid
denoted by

T

on the brane, which is the imprint of the bulk geometry through the junction
condition (2.4). Using
K
ab
= n
;
e

a
e

b
(3.5)
with a, b and , labelling the brane and bulk coordinates respectively, it is just a matter of
calculation to obtain
K
ij
=

/
a

f(a) +
a

ij
, (3.6)
K

=
(
K
2
r
2
h
a
4
+
a
a
)
_
H
2
+
K
a
2

K
2
r
2
h
a
4
, (3.7)
where is the proper time of the brane and i, j label the coordinates of the spatial section,
with H a/a is the Hubble parameter.
ij
is the metric of the unit 3-sphere. Using (3.6-3.7)
for the extrinsic curvature in (2.4) and considering

T

in a form of a perfect uid on the

brane we nd

=
s
_
1

1
2(
BH
)

s
_
, (3.8)

P =
(1 + 2)
2
+
s
( T)
3(
s
+ )
, (3.9)
where
BH
is a characteristic 3-density, proportional to the density of the bulk black hole,
averaged within our 3-brane, dened as:

BH

3
2
k
H
4
0
r
2
h
8G
N
a
4
, (3.10)
while
k
//H
2
0
. Comparing (2.13) with (3.8) we see that the integration constant from
the previous section could be interpreted as the mass of the black hole in the bulk, given in
terms of the horizon radius as
= 3[
k
[H
2
0
r
h
, (3.11)
with the comparison between (3.9) and (2.14) further indicating that = 1, and as a result,
at =
s
the pressure becomes singular. Moreover, as promised in the previous section,

2
r
2
h
< 0, which is necessary for positive energy (or ADM mass) initial conditions.
We note that
+
is non-zero, even for =
BH
= 0, which is often known as the
self-accelerating (SA) branch in the literature, as the universe can have acceleration, even
in the absence of a cosmological constant (or brane tension). However, as discussed in the
introduction, SA branch suers from a negative energy ghost instability. On the other hand,

, known as the normal branch, does not suer from the same problems, and may well
provide a healthy eective description of bulk gravity (e.g. [10]). In what follows, we outline
constraints on both branches for the sake of completeness.
In total, we have three adjustable parameters in our model:
s
, /, and r
h
. We shall
next consider the constraints on these parameters. We nd two limits on
s
. One is from
demanding reality of all quantities in (3.8), i.e.

s
2 (
BH
) . (3.12)
5

1 1
0.0 0.2 0.4 0.6 0.8 1.0
0.00
0.05
0.10
0.15
0.20
0.25
0.30

BH

Figure 1. The shaded area shows the allowed values of

s
and
BH
for both branches (pink), and
only

or the normal branch (gray). The red solid line indicates those values of
s
and
BH
for which
pressure becomes singular. We have chosen [ /[ < = 0.1 in this gure.
where the equality indicates a pressure singularity. The other comes from the fact that,
thus far, cosmological observations have not detected any eect of the induced uid , which
implies that the density of the induced matter on the brane should be small compared to
normal matter in the universe. These constraints are often expressed in terms of the transition
scale r
c
[19], where
r
c

_
3
16G
N

s
_
1/2
=
G
b
G
N
, (3.13)
which is constrained to be bigger than todays cosmological horizon scale (e.g., [14]). There-
fore, we impose a conservative bound
[ [ , (3.14)
where 1.
The constraints (3.12) and (3.14) restrict the parameter space. To clarify this we employ
equation (3.8), investigating the positive and negative branches separately. Consider rst the
positive branch. Solving (3.14) for
+
yields the upper bound

s

2
2
_
1 +

BH

_
1
, for
+
(self accelearting branch), (3.15)
which along with eq. (3.12) bounds
s
within a certain range, i.e. the pink shaded area in
gure (1). The red line in this gure shows the values for which pressure becomes singular.
Note that the lower bound (3.12) becomes important only if
BH
> ; condition (3.12) is
automatically satised for
BH
< , since
s
is always positive by denition. Both upper
and lower limits coincide at
BH
= (1 + /2); that is there are upper bounds for both

BH
(1 + /2) and
s
.
6
Figure 2. 3D plot for
k
0.01 versus log r
h
from present time (a = 1) back to Big Bang
Nucleosynthesis (a 10
10
). The red plane indicates pressure singularity while the green plane is
where r
h
= r
3
= a/

k, i.e. when our brane leaves the white hole horizon. The blue lines and the
black strip (visible at the upper right as a triangle, and continuing underneath the green surface)
dicate for a given
k
, r
h
how the radius of holographic universe evolves from BBN up to present
time; e.g. the black strip represents a holographic universe that emerges from the pressure singularity
during the radiation era, passes through the white hole horizon at a 0.01 1, and eventually is just
outside the horizon at the present time.
Considering now the negative branch

in (3.14), we obtain upper and lower limits on

s
as

s

2
2
_
1

BH

_
1
, for

BH

< 1 (

, normal branch), (3.16)

s

2
2
_
1 +

BH

_
1
, for

BH

> 1 +

2
(

, normal branch). (3.17)

This allowed region is shown in Figure (1) with gray and pink shaded areas. Note that
the red solid line representing the pressure singularity sets the lower bound for
s
/ within
(1 /2) <
BH
< (1 + /2).
So far we have found limits on
s
and
BH
. Since the value of
BH
depends on the pair

k
, r
h
, it is interesting to consider possible limits on these parameters, and how they aect
the cosmological evolution of our brane. This has been shown in a 3D plot in Figure 2. Note
that any given value for
BH
in Figure 1 corresponds to a line in the
k
, r
h
plane in Figure
2. Let us examine this gure more carefully:
First, note that the gure is plotted for the negative (or normal) branch, which, as
discussed above, is physically more relevant. The empirical upper limit for the spatial cur-
vature of the universe
k
0.01 (e.g. [20]) is indicated by the purple vertical plane in
the gure. The red surface represents those pairs of
k
, r
h
for which
BH
= 1.05(a) from
present time (log a = 0) back to Big Bang Nucleosynthesis (BBN; a 10
10
). Here we have
chosen the empirical bound 0.1 in (3.14), and have taken BBN as the earliest constraint
on deviations from the standard cosmological model. As we noted before, according to the
7
reality constraint (3.12) this is the maximum allowed value for
BH
at a given time. Therefore
the whole area under the red surface is not allowed. Moreover, the red surface also shows
the possible choices of the pairs
k
, r
h
for which the pressure becomes singular for a given
a. Consequently no pressure singularity could happen for pairs
k
, r
h
chosen to be above
the red plane at any given time.
The green plane indicates those pairs
k
, r
h
for which the radius of our 4 dimensional
universe coincides with the black hole horizon in the 5 dimensional bulk, i.e. r
h
= r
3
= a/

k.
Therefore, for any
k
, r
h
under the green plane, we have r
3
< r
h
. For those pairs
k
, r
h

chosen to be above this plane the radius of our holographic universe is larger than the horizon
radius, meaning that our present cosmos lies outside the horizon of the black hole in the bulk,
i.e. r
3
> r
h
. Subsequently, suppose we choose any pair of
k
, r
h
above the green plane
at the present time (the log a = 0 plane) and move backwards in time. Let us assume that
the universe today has its radius larger than the horizon in the bulk black hole. Moving
backwards to early times (log a = 10 plane), as the radius of the universe (proportional to
scale factor a) decreases, it may or may not cross the green plane. This has been illustrated
with the upper two blue lines in Figure 2, the lower of which pierces the green plane at some
value of r
h
near log a 5.
Indeed, crossing the green plane means that at some early time the radius of the universe
was smaller than the horizon radius. Since nothing can escape the horizon of a black hole,
one would exclude those pairs of
k
, r
h
for which their corresponding blue lines at some
a > 10
10
cross the green plane. Consequently the pairs highlighted with orange plane are
possible choices of parameters
k
, r
h
that satisfy r
3
r
h
for
k
0.01 at a = 10
10
.
Consequently, one may interpret the crossing r
3
= r
h
before BBN (0 < a < 10
10
) as
the emergence of the holographic universe out of a collapsing star: this scenario replaces
the Big Bang singularity. The overall picture of this proposal is shown in the Penrose diagram
in Figure (3)-left, which is reminiscent of the core-collapse of a supernova.
Another possibility is to consider a white hole in the bulk rather than a black hole.
With this scenario, it is possible for the universe to be inside the horizon at any time up
to the present since all matter eventually emerges from the white hole horizon. Therefore
the entire range of pairs
k
, r
h
above the red surface is allowed; the lowest blue line in
Figure 2 illustrates one such possible scenario. In this picture one may interpret the pressure
singularity as a holographic description of the Big Bang that takes place at a < 10
10
. Hence
those pairs
k
, r
h
with
k
0.01 satisfying
BH

r
(a = 10
10
), i.e. lie above the
intersection of the red surface and the a = 10
10
plane are allowed
1
. For instance, choosing
any value for
k
in the range 10
4

k
10
2
with its corresponding horizon radius,
i.e. r
h

r
/H
0

k
, represents a holographic universe that emerges from the pressure
singularity during the radiation era, passes through the white hole horizon at a 0.01 1,
and eventually is just outside the horizon at the present time. This is illustrated with a black
strip in Figure 2, visible at the upper right of the diagram and continuing underneath the
green surface toward the upper left. For any
k
< 10
4
, the universe is inside the horizon
at the present time but (given that its expansion is now dominated by the cosmological
constant), it will expand indenitely and eventually intersect the horizon in the future. The
overall picture for this scenario has been shown in the Penrose diagram in Figure (3)-right.
From the physical point of view, the former scenario, which we can dub the black
hole universe is more plausible than the latter white hole universe. The reason is that
1
We have chosen = 0.1 in (3.14) and
r
=
0

r
/a
4
.
8
BBN
black hole singularity

BBN
pressure singularity
black hole singularity
white hole singularity
Figure 3. Penrose diagram for the dynamic brane (our universe) in blue for the black hole (left) or
the white hole (right) in the bulk, where the green line indicates a collapsing shell (or star), or the
white hole horizon respectively.
the region inside a white hole horizon is to the future of a 4D white hole naked singularity
(Figure 3-right), which makes the brane dynamics, at best contrived, and at worst ill-dened.
In particular, it is hard to physically justify why this singularity (i.e. high curvature region) is
preceded by a smooth zero temperature space-time. For example, it would be in contrast
to (and thus more contrived than) the thermal bath that is the outcome of the big bang
singularity.
4 Brane Atmosphere and Cosmological Perturbations
In this section, we introduce a mechanism to generate scale-invariant cosmological pertur-
bations in our holographic big bang. As the holographic uid is sub-dominant for most
of the cosmic evolution, one expects the standard cosmological perturbation theory, that
has been extremely successful in explaining cosmic microwave background observations (e.g.
[1, 20], amongst other observational probes), to be applicable. The uid will dominate cosmic
evolution at very late times, but that can be avoided for suciently large r
c
or small a.
For super horizon perturbations, general arguments based on locality and causality
imply that one can use Friedmann equations with independent constants of motion, within
independent Hubble patches. In the presence of adiabatic perturbations, which are currently
consistent with all cosmological observations (e.g. [1, 20]), these independent Hubble patches
would only dier in their local value of comoving spatial curvature /. This is often quantied
using the Bardeen variable, , where:
/
2
3

2
, (4.1)
or equivalently the comoving gauge linearized metric takes the form
ds
2
= N
2
dt
2
+ a(t)
2
[(1 + 2)
ij
+ h
ij
] dx
i
dx
j
, (4.2)
9
where h
ij
is a traceless 3-tensor. Planck (+WMAP) observations [20] show that has a
near-scale-invariant spectrum of perturbations:
k
3
2
2
P

(k) = (2.196 0.059) 10

9
_
k
0.05 Mpc
1
_
0.03970.0073
, (4.3)
where k is the comoving wavenumber for spatial uctuations.
Given that we assumed Z
2
(or mirror) boundary conditions for our 3-brane, we can
imagine an atmosphere composed of bulk degrees of freedom, which is stratied just outside
the 3-brane, due to the gravitational pull of the black hole. Here, we argue that the ther-
mal uctuations in the atmosphere of the 3-brane induce a near-scale-invariant spectrum of
curvature perturbations ( Eq. 4.3) on our cosmological brane.
Let us rst compute the power spectrum of density uctuations for a thermal gas of
massless scalar particles in (4+1)-dimensional at spacetime. The thermal 2-point correlation
function of a free scalar eld is given in terms of the Bose-Einstein distribution:
(x)(y)
T
=
_
d
4
k
a
(2)
4
_
2
exp(/T) 1
+ 1
_
exp[ik
a
(x
a
y
a
) i(x
0
y
0
)]
2
, (4.4)
where k
a
is the spatial wave-number in 4+1D (1 a 4) , and we used E = =

k
a
k
a
for
massless particles. Now, using the denition of energy density:
(x) =
1
2

2
+
1
2

a
, (4.5)
straightforward manipulations using Eq. (4.4) yield
(x)(y)
T

5
8

_
d
4
k
a
(2)
4
_
1
exp(/T) 1
+
1
2
_
exp[ik
a
(x
a
y
a
) i(x
0
y
0
)]

2
. (4.6)
Let us next consider how these density uctuations aect metric uctuations. As a rst
attempt, we focus on the linear scalar metric uctuations in (4+1)-dimensions, which in the
longitudinal gauge can be written as:
ds
2
= (1 + 4
4
)dt
2
+ (1 2
4
)
ab
dx
a
dx
b
, (4.7)
where
4
is the analog of the Newtonian potential. We can then use 4d Poisson equation

2
4

4

8G
b
3
to nd the statistics of scalar metric uctuations. Using Eq. (4.6), we can
nd the equal-time correlator of
4
:

4
(x
a
)
4
(y
a
)
T

5
8
T
6
_
8G
b
3
_
2
_
d
4
k
(2)
4
exp[ik
a
(x
a
y
a
)]
k
4
M
_
k
T
_
, (4.8)
where
M()
_
d
4

(2)
4

_
1
2
+
1
exp(
+
) 1
_ _
1
2
+
1
exp(

) 1
_
, (4.9)

a
+
1
4

a
, (4.10)
while we have dropped the power-law UV-divergent term ( [cut-o]
6
), e.g. using dimen-
sional regularization. This UV-divergent term does not depend on temperature, and presum-
ably can be cancelled with appropriate counter-terms in other regularization schemes.
10
Now, we notice that for small k , T, we have
M()
15
R
(5)

2
+O(
2
) 1.576 +O(
2
), (4.11)
where
R
is the Riemann zeta function. Therefore, Eq. (4.8) implies that the 4d Newtonian
potential, due to thermal uctuations, has a scale-invariant power spectrum of the amplitude
of G
b
T
3
(T/M
5
)
3
.
It is easy to understand this result on dimensional grounds. Looking at the low frequency
limit T of thermal density uctuations (4.6), we notice that the argument inside the
absolute value becomes the delta function. In other words, the densities are only correlated
within a thermal wavelength T
1
, and only have white noise, or a at power spectrum, on
large scales
2
. Then, Poisson equation implies that the potential power spectrum scales as k
4
,
yielding a logarithmic real-space correlation function, or equivalently, a at dimensionless
power spectrum.
So far, all we have done is to study the uctuations of a statistically uniform 4 di-
mensional thermal bath. While the scale-invariance of this result is suggestive, it is not
immediately clear what this might imply (if anything) for cosmological curvature perturba-
tions on our 3-brane. To answer this question, we will rst assume that, at some point in
its early cosmological evolution, our 3-brane was in static equilibrium with its thermal 4d
atmosphere. Then a comparison of (4.2) and (4.7) implies that
(x
i
) =
4
(x
i
, x
4
= 0) (4.12)
assuming Z
2
boundary conditions at x
4
= 0. Note that this boundary condition modies
the thermal spectrum (4.6) within a thermal wavelength of the 3-brane; given that gravity is
a long-range force, we do not expect this to signicantly aect the long wave-length metric
uctuations. Therefore, using (4.12), we can put forth our prediction for the power spectrum
of cosmological curvature uctuations:
k
3
2
2
P

(k) =
5
32
3
_
8G
b
T
3
b
3
_
2
_

dx
(1 + x
2
)
2
M
_
k
a
b
T
b
_
1 + x
2
_
=
_
25
R
(5)
3072
4
+O
_
k
a
b
T
b
_
2
_
_
T
b
M
5
_
6

_
8.66 10
5
+O
_
k
a
b
T
b
_
2
_
_
T
b
M
5
_
6
(4.13)
where T
b
is the temperature of the bulk atmosphere, at the moment of equilibrium, where the
scale factor is a
b
. Furthermore, we used the denition of 5d Planck mass (1.2) to substitute
for G
b
. Comparing Eq. (4.13) with Eq. (4.3) gives the experimental constraint on the
(eective) temperature of the atmosphere:
T
b
M
5
= 0.17139 0.00077, (4.14)
for the comoving scale of k 0.05 Mpc
1
. While T
b
is for the atmosphere in the bulk, based
on the rate of change in spatial geometry, we may expect the de-Sitter temperature of the
2
Note that this is a general feature of Bose-Einstein distributions, on any space-time dimension.
11
boundary to set a minimum for T
b
. Therefore, we expect:
H
2
T
b
0.17 M
5
, (4.15)
The slight deviation from scale-invariance in Eq. (4.3), which is at the level of 4%, and
is now detected with Planck at > 5 level, is not predicted in our simple model of thermal
free 5D eld theory. In the next section, we will speculate on the possible origins of this
deviation in our set-up, even though we postpone a full exploration for future study.
5 Summary and Discussions
In the context of DGP brane-world gravity, we have developed a novel holographic perspec-
tive on cosmological evolution, which can circumvent a big bang singularity in our past,
and produce scale-invariant primordial curvature perturbations, consistent with modern cos-
mological observations. In this paper, we rst provided a pedagogical derivation for the
cosmological evolution of DGP braneworld in FRW symmetry from rst principles, and then
connected it to motion in the Schwarszchild bulk geometry, extending the analyses in [15] to
realistic cosmologies. Focusing on the pressure singularity uncovered in [15], we showed that
it is generically encountered at early times as matter density decays more slowly than a
4
.
However, we show that the singularity always happens inside a white hole horizon, and only
happens later than Big Bang Nucleosysntheis (BBN) for a small corner of the allowed param-
eter space (i.e. the base of black strip in Fig. 2). Therefore, it can never be created through
evolution from smooth initial conditions. This yields an alternative holographic origin for
the big bang, in which our universe emerges from the collapse a 5D star into a black hole,
reminiscent of an astrophysical core-collapse supernova (Fig. 3-left). In this scenario, there is
no big bang singularity in our causal past, and the only singularity is shielded by a black hole
horizon. Surprisingly, we found that a thermal atmosphere in equilibrium with the brane
can lead to scale-invariant curvature perturbations at the level of cosmological observations,
with little ne-tuning, i.e. if the temperature is 20% of the 5D Planck mass.
We may go further and argue that other problems in standard cosmology, traditionally
solved by ination, can also be addressed in our scenario:
1. The Horizon Problem, which refers to the uniform temperature of causally disconnected
patches, is addressed, as the star that collapsed into a 5D black hole could have had
plenty of time to reach uniform temperature across its core.
2. The Flatness Problem, which refers to the surprisingly small spatial curvature of our
universe, is addressed by assuming a large mass/energy for the 5D star, M

. The
radius of the black hole horizon, r
h
, sets the maximum spatial Ricci curvature (or
minimum radius of curvature) for our universe, and thus can only dominate at late
times. If one assumes that the initial Hubble constant is the 5D Planck mass, which
is supported by the scale of curvature perturbations above, we have
k
(M
5
r
h
)
2

M
5
/M

, which could become suciently small, for massive stars.

The curvature could of course be detectable at late times, as the Hubble constant
drops, depending on the scale of dark energy. However, a detection curvature should
generically accompany a detection of large scale anisotropy, as a generic black hole will
have a nite angular momentum, which would distort FRW symmetry on the scale of
the curvature.
12
3. The Monopole Problem refers to the absence of Grand Unied Theory (GUT) monopoles,
that should generically form (and over-close the universe) after the GUT phase transi-
tion. As we have replaced the singular big bang, with the emergence of a 4D universe at
a nite size, the plasma temperature never reaches GUT scale, and thus the GUT phase
transition will never have happened in the thermal history of the universe, preventing
copious production of monopoles.
To see this, we can translate the observational constraints on the DGP cosmology
(normal branch), r
c
3H
1
0
[14] into an upper limit on 5D Planck mass:
H M
5

_
H
0
M
2
4
6
_
1/3
9 MeV, (5.1)
where we used the inequality in Eq. (4.15) to bound the Hubble constant. Correspond-
ingly, the upper limit on the temperature comes from the Friedmann equation in the
radiation era, for g

species:
T
_
M
4
H
g
2

_
1/4
3 10
4
_
g

100
_
1/4
TeV T
GUT
10
12
TeV. (5.2)
Yet another attractive feature of our construction is that it lives in an asymptotically
at space-time. This potentially allows for an S-matrix description of this cosmology, through
collapse of an ingoing shell, and emergence of outcoming D-brane. This might be a promising
avenue, especially in light of signicant recent progress in understanding scattering ampli-
tudes in supergravity (e.g. [21, 22]. Furthermore, with trivial modication, this model could
also be embedded in an AdS bulk, which can potentially allow a study of the strongly-coupled
dynamics of emergence through AdS/CFT correspondence. Since embedding our braneworld
in a large AdS space-time (instead of Minkowski) simply amounts to adding a small constant
to the right hand side of e.g., Eqs. (2.6) or (2.12), it need not signicantly change any of the
quantitative results that we have found here.
Let us now comment on (some) potential problems. Perhaps the most notable problem
with the DGP model might be the claim [23] that superluminal propagation around non-
trivial backgrounds in DGP model hinders causal evolution, and UV analyticity/completion.
However this violation of causality is only a pathology for spacetimes that dont admit a
consistent chronology for (super)luminal signals [24]. Such spacetimes, e.g. Godel metric,
even exist in General Relativity, and simply point out the absence of global causal evolution in
those backgrounds. Therefore such geometries cannot emerge out of classical causal evolution.
The second objection is more subtle, and relies on the analyticity properties of the scattering
amplitudes for the DGP scalar. However, these conditions (e.g. the Froissart bound) may
be violated in the presence of massless bulk gravitons. Therefore, these arguments would
leave the door open for a possible UV completion via e.g. string theory and/or AdS/CFT
correspondence.
Another possible pathology of the DGP model is copious spontaneous production of
self-accelerating branes in the bulk [15], which is estimated via Euclidean instanton methods.
However one may argue that, since self-accelerating branches have catastrophic ghost insta-
bilities, they should be excised (or exorcised) from the Hilbert space of the system. Given
that one cannot classically transition from the normal branch to the self-accelerating branch,
this modication would not aect the semi-classical behavior, but would prevent tunnelling
into unphysical states.
13
Finally, let us comment on potential testability of this model. As we pointed out, the
simple model of cosmological perturbations, developed in Sec. 4 is already ruled out by cosmo-
logical observations at > 5 level, as it does not predict any deviations from scale-invariance.
However, it is easy to imagine small corrections that could lead to a 4% deviation from
scale-invariance, especially given that bulk temperature is so close (i.e. 20% of) the 5D
Planck temperature. In the context of our model, the red tilt of the cosmological power
spectrum implies that the amplitude of 5D bulk graviton propagator, which enters in Eq.
(4.8), is getting stronger in the IR, suggesting gradual unfreezing of additional polarizations
of graviton. For example, this is what one would expect in cascading gravity [25], where
DGP bulk is replaced by a 4-brane, which is itself embedded in a 6D bulk. Similar to the
ordinary DGP, the transition in at space happens on length-scales larger than M
3
5
/M
4
6
, as
the scalar eld associated with the motion of the 4-brane in the 6D bulk becomes weakly
coupled, and boosts the strength of the gravitational exchange amplitude.
A related issue is that the gravitational Jeans instability of the thermal atmosphere
kicks in for k < k
J
0.2 T
b
(T
b
/M
5
)
3/2
10
2
T
b
, which may appear to limit the range
of scale-invariant power spectrum to less than the current observations. However, the time-
scale for the Jeans instability can be signicantly longer than the Hubble time, thus limiting
its maximum growth. Nevertheless, one may consider the residual Jeans instability as a
potential origin for the slight red tilt (i.e. n
s
< 1) of the observed power spectrum. We
defer a consistent inclusion of gravitational backreaction on the 5d thermal power spectrum
(which should account for the impact of Jeans instability) to a future study.
We should stress that, at this point, the development of a mechanism responsible for the
observed deviation from scale-invariance is the most immediate phenomenological challenge
for our scenario. The next challenge would be a study of the interactions that lead to devia-
tions from scale-invariance, and whether they satisfy the stringent observational bounds on
primordial non-gaussianity [26]. Other interesting questions might be, given that the emer-
gence from the 5D black hole might happen at relatively low temperatures, could there be
observable predictions for gravitational waves (either on cosmological scales, or for gravita-
tional wave interferometers), or even modications of light element abundances in Big Bang
Nucleosynthesis.
Ultimately, an entire new world might emerge Out of the White Hole, and replace
Big Bang with a mere mirage of a non-existent past!
Acknowledgements We would like to thank Nima Doroud, Gregory Gabadadze, Ruth
Gregory, Rob Myers, Paul McFadden, Eric Poisson, Kostas Skenderis, Misha Smolkin, Marika
Taylor and Herman Verlinde for inspiring discussions and useful comments. This work was
supported by the Natural Science and Engineering Research Council of Canada, the Univer-
sity of Waterloo and by Perimeter Institute for Theoretical Physics. Research at Perimeter
Institute is supported by the Government of Canada through Industry Canada and by the
Province of Ontario through the Ministry of Research & Innovation.
References
[1] Planck Collaboration Collaboration, P. Ade et al., Planck 2013 results. I. Overview of
products and scientic results, arXiv:1303.5062.
[2] L. Randall and R. Sundrum, A Large mass hierarchy from a small extra dimension,
Phys.Rev.Lett. 83 (1999) 33703373, [hep-ph/9905221].
14
[3] H. Collins and B. Holdom, Brane cosmologies without orbifolds, Phys.Rev. D62 (2000) 105009,
[hep-ph/0003173].
[4] G. Dvali, G. Gabadadze, and M. Porrati, 4-D gravity on a brane in 5-D Minkowski space,
Phys.Lett. B485 (2000) 208214, [hep-th/0005016].
[5] A. Vainshtein, To the problem of nonvanishing gravitation mass, Phys.Lett. B39 (1972)
393394.
[6] I. Savonije and E. P. Verlinde, CFT and entropy on the brane, Phys.Lett. B507 (2001)
305311, [hep-th/0102042].
[7] S. S. Gubser, AdS / CFT and gravity, Phys.Rev. D63 (2001) 084017, [hep-th/9912001].
[8] L. Pilo, R. Rattazzi, and A. Zaaroni, The Fate of the radion in models with metastable
graviton, JHEP 0007 (2000) 056, [hep-th/0004028].
[9] M. A. Luty, M. Porrati, and R. Rattazzi, Strong interactions and stability in the DGP model,
JHEP 0309 (2003) 029, [hep-th/0303116].
[10] A. Nicolis and R. Rattazzi, Classical and quantum consistency of the DGP model, JHEP 0406
(2004) 059, [hep-th/0404159].
[11] C. Charmousis, R. Gregory, N. Kaloper, and A. Padilla, DGP Specteroscopy, JHEP 0610
(2006) 066, [hep-th/0604086].
[12] N. Kaloper, Gravitational shock waves and their scattering in brane-induced gravity, Phys.Rev.
D71 (2005) 086003, [hep-th/0502035].
[13] W. Fang, S. Wang, W. Hu, Z. Haiman, L. Hui, et al., Challenges to the DGP Model from
Horizon-Scale Growth and Geometry, Phys.Rev. D78 (2008) 103509, [arXiv:0808.2208].
[14] T. Azizi, M. Sadegh Movahed, and K. Nozari, Observational Constraints on the Normal
Branch of a Warped DGP Cosmology, New Astron. 17 (2012) 424432, [arXiv:1111.3195].
[15] R. Gregory, N. Kaloper, R. C. Myers, and A. Padilla, A New perspective on DGP gravity,
JHEP 0710 (2007) 069, [arXiv:0707.2666].
[16] J. D. Brown and J. York, James W., Quasilocal energy and conserved charges derived from the
gravitational action, Phys.Rev. D47 (1993) 14071419, [gr-qc/9209012].
[17] K.-i. Maeda, S. Mizuno, and T. Torii, Eective gravitational equations on brane world with
induced gravity, Phys.Rev. D68 (2003) 024033, [gr-qc/0303039].
[18] L. Susskind, The World as a hologram, J.Math.Phys. 36 (1995) 63776396, [hep-th/9409089].
[19] C. Deayet, Cosmology on a brane in Minkowski bulk, Phys.Lett. B502 (2001) 199208,
[hep-th/0010186].
[20] Planck Collaboration Collaboration, P. Ade et al., Planck 2013 results. XVI. Cosmological
parameters, arXiv:1303.5076.
[21] F. Cachazo and Y. Geyer, A Twistor String Inspired Formula For Tree-Level Scattering
Amplitudes in N=8 SUGRA, arXiv:1206.6511.
[22] F. Cachazo, L. Mason, and D. Skinner, Gravity in Twistor Space and its Grassmannian
Formulation, arXiv:1207.4712.
[23] A. Adams, N. Arkani-Hamed, S. Dubovsky, A. Nicolis, and R. Rattazzi, Causality, analyticity
and an IR obstruction to UV completion, JHEP 0610 (2006) 014, [hep-th/0602178].
[24] J.-P. Bruneton, On causality and superluminal behavior in classical eld theories: Applications
to k-essence theories and MOND-like theories of gravity, Phys.Rev. D75 (2007) 085013,
[gr-qc/0607055].
[25] C. de Rham, G. Dvali, S. Hofmann, J. Khoury, O. Pujolas, et al., Cascading gravity: Extending
15
the Dvali-Gabadadze-Porrati model to higher dimension, Phys.Rev.Lett. 100 (2008) 251603,
[arXiv:0711.2072].
[26] Planck Collaboration Collaboration, P. Ade et al., Planck 2013 Results. XXIV. Constraints
on primordial non-Gaussianity, arXiv:1303.5084.
16